jamsocket-jupyter-notebook tf

Base repo for running Jupyter Notebooks on the Jamsocket platform

Deploying

1. Log in to Jamsocket and create a new service called jupyter-notebook. (Sign up for a Jamsocket account if you don’t already have one. You can do everything listed here on the free plan.)
2. Reach out to us on Discord or by email to ask us to enable Jamsocket compatibility mode for that service. (This will soon be exposed directly in the web UI, but it's a manual step for now.)
3. Fork this repo as jupyter-notebook to your own GitHub account. Note that the name of the repo should match the name of the service, but you can use another name as long as they both match.
4. Set your Jamsocket account name as the JAMSOCKET_ACCOUNT value in your GitHub secrets.
5. Generate an access token at app.jamsocket.com/settings and set it as the JAMSOCKET_TOKEN value in your GitHub secrets.
6. Push your repo to your GitHub account, and your Jupyter Notebook container will be automatically built and pushed to the Jamsocket service you selected.
7. Use the Jamsocket API to spawn instances of your notebook. It will return a URL that you can use to access the notebook.

Customizing

You can customize the initial notebook(s) and Python packages as follows:

1. Add your notebook(s) to the /notebook directory.
2. Add your notebook's python dependencies to requirements.txt (alternatively, if you prefer pipenv, add them with pipenv install and then run pipenv run freeze).
3. Replace /notebooks/hello-jamsocket.ipynb in Dockerfile with the path to the notebook you'd like to be automatically loaded. If you want the user to see a file listing of the /notebooks directory by default, remove that line entirely.
4. Run ./freeze.sh to create a freeze.txt file from requirements.txt. This pins the version of all of the dependencies of the libraries in requirements.txt.

To run the jupyter-notebook locally for testing:

docker build -t jupyter-notebook .
docker run -p 8080:8080 --env PORT=8080 -it jupyter-notebook
open localhost:8080

Jupyter Lab

By default, this runs the jupyter notebook command. If you want to run the jupyter lab command instead, you can set the JUPYTER_SUBCOMMAND environment variable to lab.